What is artificial intelligence? Types, uses and Saudi examples

Artificial intelligence (AI) is software that learns from data to do tasks that normally need human intelligence: recognising images, understanding language, predicting demand, recommending a choice or drafting a text. In Saudi Arabia it already reads documents, answers customers in Arabic, forecasts sales and flags suspicious payments, and it sits at the centre of the national data and AI strategy.

This guide explains what artificial intelligence is in plain terms, how it learns, the main types you will hear about, where it is used in the Kingdom and why it matters for companies. Business examples are marked as illustrative; figures come from the official and primary sources listed at the end, opened on 14 October 2026. O AI is not affiliated with OpenAI or any other company named here.

What is artificial intelligence? Three definitions that agree

There is no single legal definition, but the main ones say the same thing in different words.

• OECD: an AI system is a machine-based system that, for explicit or implicit objectives, infers from the input it receives how to generate outputs such as predictions, content, recommendations or decisions that can influence physical or virtual environments. The OECD adds that AI systems vary in how autonomous and adaptive they are after deployment [1].
• SDAIA's AI Adoption Framework (2025): software systems that rely on advanced technologies to predict, generate content, provide recommendations or make decisions, with varying levels of autonomy, depending on the data and the context in which they operate [2].
• SDAIA's AI Ethics Principles: a collection of technologies that can enable a machine or system to sense, comprehend, act and learn [3].

Put simply, an AI system takes an input (a photo, a sentence, a column of sales figures), works out a pattern it learned from earlier data, and returns an output (a label, a forecast, a suggestion or a new piece of text).

The idea is older than most people think. The term comes from a proposal dated 31 August 1955 by John McCarthy, Marvin Minsky, Nathaniel Rochester and Claude Shannon for a summer research project at Dartmouth College. It rested on the conjecture that every aspect of learning, or any other feature of intelligence, can in principle be described so precisely that a machine can be made to simulate it [4].

How does AI work? Rules versus learning

Ordinary software follows rules a programmer wrote: "if the invoice total is above SAR 10,000, send it to the finance manager." AI, and machine learning in particular, works the other way round. You give the system many examples and it learns the rule itself. SDAIA's framework describes machine learning as technology that lets systems learn from data and improve their performance without being explicitly programmed [2].

SDAIA lists three forms of machine learning [2]:

• Supervised learning: learning from labelled examples, such as thousands of past invoices marked "approved" or "rejected".
• Unsupervised learning: finding groups in unlabelled data, such as customers who buy in similar ways.
• Reinforcement learning: learning by trial and error from feedback, as systems that optimise delivery routes or energy use can do.

Because the system learns from examples, the examples decide much of the result. Data that is old, one-sided or full of errors produces a model with the same faults, which is why data preparation takes so much of the work in any serious AI project.

Types of artificial intelligence

You will meet two ways of classifying AI. One describes how capable a system is; the other describes the technique it uses.

By capability: narrow, general and super AI

IBM's widely used classification separates three levels [5]:

• Narrow AI (weak AI): systems that do one task or a set of related tasks. It is the only type of AI that exists today, from a spam filter to the most advanced chat assistant.
• General AI (strong AI): a system that could learn and apply knowledge across any task like a person. It remains a theoretical concept.
• Super AI: intelligence beyond human ability in every field. It is strictly theoretical.

So when a product claims to "think like a human", treat it as marketing. Every system you can buy is narrow AI, however broad it looks.

By technique: the branches you will meet at work

SDAIA's framework names the technologies that make up modern AI [2]. The table pairs each one with an illustrative Saudi business example.

Generative AI tools such as ChatGPT, Gemini and HUMAIN Chat are the branch most people now meet first. AI agents, software that chooses its own next step and uses other systems to complete a task, build on the same models; our guide to AI agents, RPA and workflow tools explains when a business needs one.

How AI, machine learning, deep learning and generative AI fit together

Think of nested circles. Artificial intelligence is the outer circle. Machine learning is the large part of it that learns from data. Deep learning is the part of machine learning that uses many-layered neural networks, and today's generative AI models are built with deep learning. Not every AI system learns, either: a hand-built expert system that follows rules written by specialists is AI too.

Uses of artificial intelligence in Saudi Arabia

The General Authority for Statistics (GASTAT) found that 33.1% of establishments used AI technologies in 2025, up 20% on 2024, according to its Establishments ICT Access and Usage Survey as reported by Arab News [6]. Use was highest in information and communications (61.1%), financial and insurance activities (52.9%) and education (51%), followed by transportation and storage (44.4%) and professional activities (43.9%) [6].

National projects show the Kingdom building its own capability as well as buying it:

• ALLaM: a series of Arabic and English language models developed by the National Center for Artificial Intelligence at SDAIA; one 7-billion-parameter version is published openly [7].
• HUMAIN Chat: an Arabic chat app launched on 25 August 2025 by HUMAIN, a company owned by the Public Investment Fund. It runs on the ALLaM 34B model, supports several Arabic dialects and, according to Saudipedia, is hosted on HUMAIN's infrastructure inside the Kingdom [8].

Illustrative examples. The uses below are hypothetical; they show how the same techniques appear in ordinary Saudi businesses.

In each case, AI handles the repetitive reading, sorting or forecasting, and a person keeps the decision that carries consequences.

Why artificial intelligence matters

For the Kingdom. The National Strategy for Data and AI sets targets that include ranking among the top 15 countries in AI, building a workforce of more than 20,000 data and AI specialists and attracting about SAR 75 billion of investment in data and AI, in phases that run to 2030 [9]. SDAIA reports that 66 of the 96 objectives of Vision 2030 relate directly or indirectly to data and AI [10].

For companies. Stanford's 2026 AI Index reports that organisational adoption of AI reached 88% [11]. SDAIA's framework lists what entities gain: automating routine administrative work, shorter processing times, better use of resources by predicting risks and needs, and decisions supported by data from several sources [2]. For a small business, the practical gain is usually hours returned to staff on reading, typing and sorting.

For people. The same AI Index reports that 4 in 5 university students now use generative AI [11]. Knowing what these tools can and cannot do is becoming a basic work skill, like using a spreadsheet.

What AI cannot do, and the rules that apply in Saudi Arabia

AI is powerful within limits, and the limits matter as much as the capabilities.

• It does not understand like a person. All available AI is narrow AI [5]; it finds patterns, and a system that is excellent at one task can fail badly at a neighbouring one.
• It can be confidently wrong. SDAIA warns that generative models may produce "facts" that are complete fiction, and that users are responsible for verifying generated content [12].
• It repeats the faults in its data. SDAIA's ethics principles ask that training data be checked for bias and be representative of the people affected [3].

SDAIA's AI Ethics Principles apply to all segments of society in the Kingdom, including public, private and non-profit entities [12]. They set seven principles: fairness; privacy and security; humanity; social and environmental benefits; reliability and safety; transparency and explainability; and accountability and responsibility [3]. They also sort AI systems by risk [3]:

• Little or no risk, such as spam filters: no restrictions, though ethical compliance is recommended.
• Limited risk: the ethics principles apply.
• High risk to basic rights: assessments before and after deployment, plus the relevant statutory requirements.
• Unacceptable risk, such as social profiling or exploiting children: not allowed.

Personal data used by AI also falls under the Personal Data Protection Law. This section is general information; for a specific case, check SDAIA's guidance or ask a licensed advisor. If you want to gauge how ready your company is, our guide to the SDAIA AI Adoption Framework translates it into steps for smaller firms.

Your next step

Write down three tasks in your company that involve a lot of reading, typing, sorting or forecasting, and note how many hours each takes per week. That list, not a vendor's demo, is where artificial intelligence earns its place. Our guide on how AI can help your business shows how to pick one task and test it in a short trial before you spend. You can also read about O AI, a software and AI company in Al Khobar, on our home page.